How does the memory bandwidth of PCIex16 2.0 does not bottleneck a GTX 970?

G

Guest

Guest
Memory bandwidth of PCIe16 2.0 is 500MB/s where as memory bandwidth of GTX 970 is upto 196GB/s. So how come this great difference in bandwidth doesn't bottleneck it. I even heard that no single card can be bottlenecked by a PCIex16 2.0 slot. Please enlighten me about this topic.
 
Solution
1) It's 500mb/s PER LANE, and you have 16 of them, so 8GB/s
2) Texture compression helps out a ton
3) 99% of the data from one frame is reused in the next frame, so the actual data transfer between card and CPU is actually much smaller than you think. That's why there's a 4GB memory on that 970, so that you can keep data that you'll reuse or need later!
1) It's 500mb/s PER LANE, and you have 16 of them, so 8GB/s
2) Texture compression helps out a ton
3) 99% of the data from one frame is reused in the next frame, so the actual data transfer between card and CPU is actually much smaller than you think. That's why there's a 4GB memory on that 970, so that you can keep data that you'll reuse or need later!
 
Solution
G

Guest

Guest


Thanks a lot buddy, that cleared up the confusion!!!